Credit Scores, Social Capital, and Stock Market
Participation
Jesse Bricker
Geng Li
Federal Reserve Board
∗
Federal Reserve Board
December 2016
∗
We thank our colleagues at the Federal Reserve Board and participants at the fall HFCN meeting at
the European Central Bank for helpful discussions. The assistance of the American Red Cross in providing
blood donation data is gratefully acknowledged. Jessica Hayes and Taha Ahsin provided excellent research
assistance. The views expressed herein are those of the authors and do not necessarily reflect those of the
Board of Governors of the Federal Reserve System or its staff.
Credit Scores, Social Capital, and Stock Market Participation
Abstract
While a rapidly growing body of research underscores the influence of social capital
on financial decisions and economic developments, objective data-based measurements
of social capital are lacking. We introduce average credit scores as an indicator of a
community’s social capital and present evidence that this measure is consistent with,
but richer and more robust than, those used in the existing literature, such as electoral
participation, blood donations, and survey-based measures. Merging unique proprietary credit score data with two nationwide representative household surveys, we show
that households residing in communities with higher social capital are more likely to
invest in stocks, even after controlling for a rich set of socioeconomic, preferential,
neighborhood, and demographic characteristics. Notably, such a relationship is robustly observed only when social capital is measured using community average credit
scores. Consistent with the notion that social capital and trust promote stock investment, we find the following: first, the association between average credit score and
stock ownership is more pronounced among the lower educated; second, social capital
levels of the county where one grew up appear to have a lasting influence on future
stock investment; and third, investors who did not own stocks before have a greater
chance of entering the stock market a few years after they relocate to higher-score
communities.
Keywords: Trust, Social capital, Stock market participation, Credit scores
JEL: D14, G10, O16
1
1
Introduction
The past quarter century witnessed a renaissance of research on social capital and trust.
Since the seminal work of Putnam (1993), the influence of social capital and trust has
been underscored in explaining economic outcomes in various contexts, such as economic
growth and development (Knack and Keefer, 1997), performance of institutions and judiciary
efficiency (Fukuyama, 1995; La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997), and the
rise of public schools (Goldin and Katz, 1999).1 More recently, Guiso, Sapienza, and Zingales
(henceforth GSZ) document that greater social capital promotes financial developments (GSZ
2004).
While there is broad recognition that social capital influences financial and economic activities, concrete, objective, and data-based measures of social capital have remained elusive.
Indeed, as Putnam (1995) famously wrote, “Since trust is so central to the theory of social
capital, it would be desirable to have strong behavioral indicators of trends in social trust
and misanthropy. I have discovered no such behavioral measures.” In the existing literature, social capital is often approximated by the level of trust in a society, which in turn is
often measured using self-reported trusting attitudes in household surveys.2 However, the
proper interpretation of responses to such survey questions remains a subject of active debate (Glaeser, Laibson, Scheinkman, and Soutter, 2000; Fehr, Fischbacher, von Rosenbladt,
Schupp, and Wagner, 2003; Karlan, 2005; Sapienza, Toldra-Simats, and Zingales, 2013). In
this regard, GSZ (2004) propose electoral participation and blood donations as measures
of social capital. However, these measures have been primarily applied to European countries, whereas empirical studies of how social capital may affect U.S. financial markets and
household financial decisions remain largely a void.
In this paper, we make two contributions to the measurement of social capital and its
1
Arrow (1969, 1972) was among the earliest authors who noted that trust was essential to commercial
transactions.
2
These questions are typically framed as “Generally speaking, would you say that most people can be
trusted or that you have to be very careful in dealing with people?”
2
effect on household financial decisions. First, we propose a novel measure of social capital
and trust—the average credit score of the residents of a community. Second, we examine
how social capital and trust may influence U.S. household investors’ financial decisions, with
a particular focus on their stock investment.
Credit scores are designed to predict consumers’ future default risks from their financial
debt and are estimated using rich credit and payment history data. A community’s average
credit score can serve as a good indicator of its level of social trust and social capital for
two reasons. First, credit scores reflect people’s previous experiences with credit markets
and personal finance. Many recent studies have revealed that people’s perceptions and
expectations are heavily influenced by their own past experiences. For example, Malmendier
and Nagel (2011) document that investors who experienced prolonged periods of low stock
returns are less likely to later invest in stocks. Malmendier and Nagel (forthcoming) show
past experiences with inflation strongly influence subsequent inflation expectations. The
neighborhoods with lower average credit scores tend to be economically downtrodden or hit
particularly hard during financially hard times. Arguably, residents in such communities can
feel more distrustful toward financial markets in general, including the stock market.
Second, credit scores, to a certain extent, may reveal an individual’s underlying trustworthiness beyond the likelihood of defaulting on financial obligations. For example, Dokko,
Li, and Hayes (2015) show that individuals’ credit scores are strong predictors of personal
relationship outcomes, even after controlling for credit market experiences. Thus, a neighborhood with higher average credit scores may also have higher levels of trustworthiness
and social capital. Furthermore, an individual living in such a community has a greater
opportunity to interact with more trustworthy people, which could make her more trustful
herself.
We use a large proprietary dataset—the Federal Reserve Bank of New York Consumer
Credit Panel/Equifax data—to estimate community average credit scores. We then examine
the validity of average credit scores as an indicator for social capital and contrast its merit
3
with other social capital measures in a number of statistical exercises. Specifically, we show
that, controlling for an extensive array of observable community characteristics, the average
community credit score is positively correlated with a survey-measured level of social trust
(estimated using the Social Capital Community Benchmark Survey), the general election
participation rate, and the quantity of blood donationsp; it is also negatively correlated with
the rate of consumer complaints filed with the Federal Communications Commission (FCC),
a possible measure of social distrust. Principal component analysis reveals that the average
credit score has the highest correlation with the latent factor derived from all social capital
indicators. Furthermore, a unique advantage of our proposed social capital indicator is that
the average credit score can be estimated for small geographies, such as a census tract or
even a street block, which is particularly appealing because the effects of social capital are
likely more pronounced in more granularly-defined communities.
Our analysis on stock ownership contributes to a long line of research on the topic. The
low level of stock market participation relative to what theory predicts has long been a puzzling phenomenon (Haliassos and Bertaut, 1995). Fewer than 25 percent of U.S. households
directly own stocks (Bricker and others, 2014) even though stock investments earn a substantially higher risk-adjusted return than safer assets in the long run.3 Numerous theories, such
as participation costs (Vissing-Jorgensen, 2002; Briggs, Cesarini, Lindqvist, and Östling,
2016), information barriers (Hong, Kubik, and Stein, 2004; Li, 2014), and certain behavioral
biases (Haliassos and Bertaut, 1995; Malmendier and Nagel, 2011) have been proposed to
account for the lack of stock market participation.
Our analysis, however, is intimately related to a recent strand of literature that underscores the influence of social trust on stock investment (GSZ 2008, El-Attar and Poschke,
2011; Georgarakos and Pasini, 2011). Notably, GSZ (2008) argue that as the perceived probability of being cheated increases, stock investment becomes less likely, whereas areas that
3
Only about 50 percent own stocks even after including indirect equity ownership through retirement
accounts. Throughout this paper we follow the literature and exclude indirect ownership (Brown, Ivković,
Smith, and Weisbenner, 2008; Balloch, Nicolae, and Philip, 2015; GSZ, 2004, 2008).
4
have more social capital and display greater trust tend to have higher stock market participation. They find support for this hypothesis using individual level data of self-reported trust
and stock ownership of Dutch households and cross-county evidence using the World Values
Survey. More recently, Giannetti and Wang (2016) document that, in household survey data
similar to what we use, household stock market participation decreases after the revelation
of corporate fraud in the state of their residence.
While our empirical strategy shares many similarities with GSZ (2004), we extend and
complement their work in several important aspects. First, instead of examining how selfreported trust affects stock ownership, we link the level of social capital of the community
where an investor resides with her stock investment decisions.4 Second, we study a broad
array of measures of social capital, including the one we introduce here—the community’s
average credit score—and those adopted in the existing literature—such as electoral participation and blood donations—in order to examine the consistency among these indicators
and assess their relative merit in accounting for variations in propensities of investing in
stocks. Third, to the best of our knowledge, this paper is the first that studies the relationship between social capital and stock market participation at individual investor level using
data from large U.S. household surveys.5
Specifically, we link household balance sheet information in the Survey of Consumer
Finances (SCF) and the Panel Study of Income Dynamics (PSID) to various indicators of
trust and social capital. We find that consumers residing in areas with higher average credit
scores are more likely to own stocks and to invest a greater share of their portfolio in stocks.
For example, our baseline SCF analysis indicates that investors living in a census tract
4
As a robustness check, we also use an assessment of the trusting behavior of the household provided
by survey interviewers. We find a similar positive correlation between household trust and stock market
participation as in GSZ (2008).
5
Balloch, Nicolae, and Philip (2015) and Duarte, Siegel, and Young (2012) are the only related studies
of U.S. investors we find. Balloch, Nicolae, and Philip (2015) use an Internet panel that is likely not
representative of the U.S. population as nearly 70 percent of the panel’s respondents are stock owners, many
times higher than observed in representative household surveys and administrative tax data. Duarte, Siegel,
and Young (2012) study peer-to-peer lending, which is still only used by a small, select subpopulation of
U.S. households.
5
with a one standard deviation higher average credit score have a 7 percentage point higher
probability of owning stocks. Conditional on owning stocks, the share of stocks in financial
investment portfolio is about 60 percent higher.6 Such a relationship holds against a rich set
of socioeconomic, preferential, neighborhood, and demographic characteristics. In particular,
our estimates are robust to controlling for neighborhood average stock ownership, suggesting
that it is a factor that goes beyond the information sharing channel that was also shown
in earlier studies to have a positive effect on stock investment (Hong, Kubik, and Stein,
2004; and Li, 2014). Interestingly, community electoral participation and blood donations
are only weakly associated with the likelihood of stock investment. In a horse-race-style
analysis in which household financial decisions are regressed on all social capital indicators
at our disposal, average credit scores stand out as the winner for having the most sizable
and significant estimated coefficient.
To examine if such a relationship is causal, we first show that the association between
the propensity to own stocks and community average credit score is stronger for investors
with lower educational attainment and weaker among the college educated, a finding similar
to that in GSZ (2004, 2008). We also show that the social capital level of the county where
an investor grew up appears to have a lasting influence on her future stock ownership even
years after moving out of that county. Furthermore, we leverage the longitudinal structure
of the PSID data and study stock investment dynamics. We find that investors who did
not own stocks previously have a greater chance of entering the stock market a few years
after they relocate to higher-score communities relative to comparable investors who did not
move. This trend is consistent with the narrative that relocating to a high-score community
allows the investor to interact with more trustworthy individuals and thereby develop a more
trusting attitude himself. Our exercise focuses on stock market entries that are subsequent,
instead of simultaneous, to relocation, thereby circumventing the endogeneity concerns to a
certain extent.
6
The average share of stock investment in financial portfolio is 11 to 16 percent in our samples.
6
The remainder of the paper proceeds as follows. Section 2 briefly discusses the theoretical
background of social capital and introduces the community average credit score as one of
its measures. Section 3 describes various data sources used in the paper and presents key
summary statistics. Section 4 compares average credit scores with other measures of social
capital. Sections 5 and 6 present static and dynamic analyses, respectively, of the relationship
between average credit scores and stock investment. Section 7 concludes.
2
Conceptual Framework and Related Literature
2.1
Trust and Social Capital
According to a widely cited definition from Putnam (1993), social capital is the “features of
social life, networks, norms and trust that enable participants to act together more effectively
to pursue shared objectives.” Accordingly, societies with greater social capital tend to be
more trusting, and more trusting societies are able to have stronger social connections,
positive social norms, and lower transactional costs in economic activities. A voluminous
literature has studied cross-country variations in trust and social capital, and how they help
explain differences in economic growth (Putnam, 1993; Fukuyama, 1995; Knack and Keefer,
1997; La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997; Algan and Cahuc, 2010). Our
paper will follow this tradition and interpret trust as a crucial component of social capital.
Trust also functions as a pillar of financial interactions. Arrow (1969) famously states
that “it is useful for individuals to have some trust in each other’s word. In the absence of
trust it would become very costly to arrange for alternative sanctions and guarantees, and
many opportunities for mutually beneficial cooperation would be foregone.” More recently,
GSZ (2004) study the diverging growth paths of the north and south of Italy (a celebrated
example of social capital and economic development). Both areas had similar levels of
economic development in the late 1800s, but the north of Italy had stronger social cohesion,
trust in institutions, and social capital. GSZ document that by the 20th century, areas of
7
Italy with higher social capital were more likely to develop financially and had residents that
were more likely to participate in financial equity markets. They also show that the higher
levels of social capital in the north led to stronger use of financial institutions and economic
growth.
In a similar spirit, GSZ (2008) demonstrate that lack of trust potentially contributes
to the low rate of participation in stock markets. They show that individuals who trust
others are 50 percent more likely to invest in stocks, and among stock investors, those who
trust others have a 15 percent higher stock allocation than the mean. As GSZ point out,
stock investment involves an investor sending assets to a financial intermediary (a broker,
for example), who then invests the money in the market. If invested properly the broker
makes money for both herself and the investor, and this cooperative endeavor benefits both
parties. However, if the subjective probability of being cheated in this market increases,
cooperation between market participants becomes less likely. Therefore, investing in stocks
crucially depends on trust and cooperation.7
The experience of a community’s residents, including their interactions with fellow members of the community, may shape their perception of and attitude towards the market and
the society to a great extent, and thereby influence the level of trust and social capital of the
community. For example, Malmendier and Nagel (2011, forthcoming) show that individuals’ stock investment and inflation expectations are significantly influenced by their previous
stock return and inflation experiences. In general, as one encounters more cheaters or feels
unfairly treated in personal interactions and business transactions, her trust of the society
will be undermined. Indeed, recent research has found that individuals with low trust are
more likely to have experienced a recent traumatic event or to belong to a group that has
traditionally experienced discrimination (Alesina and La Ferrrara, 2002). In addition, banking crises also have persistent negative effects on measured individual trust, especially trust
7
Trust in the financial intermediary can also increase an investor’s willingness to take risks (Gennaioli,
Schleifer, and Vishny, 2015), though this is an avenue that we do not explore in this paper.
8
in social institutions (Graeber and Zimmerman, 2016). In a similar vein, the composition of
a community can affect the trust levels therein. On the one hand, exposure to more trustworthy people may help others overcome adversity, especially those with low social standing
(Helliwell, Huang, and Wang, 2016). On the other hand, lower levels of trust are observed
among those exposed to less-trustworthy people or to conflicting social norms (Alesina and
La Ferrrara, 2002).
In spite of the increasing appreciation of the role of social capital in financial and economic developments, its measurement has been elusive to students of the subject. In the
existing literature, trust, a central element of social capital, is routinely measured with survey responses, which by design are only available for a small sample of individuals. More
recently, GSZ (2004) propose electoral participation and blood donations as social capital
measures. However, the implications of these measures of social capital have been studied
primarily in European countries and have not been systematically explored using U.S. data.
Moreover, the availability of these statistics is often restricted to a certain geographical levels. The search for an objective behavioral-based measure of social capital remains a high
research priority.
2.2
Community Average Credit Scores as an Indicator of Social
Capital
With this motivation, we propose using the average credit score in a community as an indicator of social capital in that community. Credit scores are designed to evaluate credit quality
and predict the default risk of potential borrowers. Generally, borrowers with higher credit
scores are deemed more creditworthy and have, on average, lower default rates. Broadly
speaking, a person’s default probability is affected by her willingness to repay her debt,
which we denote with ω, and her ability to repay, which we denote with η. It is therefore
helpful to represent one’s credit score as a (noisy) indicator of her default probability such
9
that
score = f (ω, η) + µ,
(1)
where µ is an error term.
Historically, lenders have long recognized a borrower’s general trustworthiness and personality as important factors influencing her debt payment history and default probability.
For example, Dokko, Li, and Hayes (2015) report that the credit reports garnered in the
1930s, in addition to debt repayment history and other financial data, collected information on borrowers’ characteristics, reputation, habits, morals, and even illegal liquor traffic
activities.8
Although modern credit reports no longer collect such soft information on borrowers’ personality and characteristics, debt payment history remains the most important determinant
of one’s credit score.9 As a result, one’s credit score contains signals that potentially reveal
underlying trustworthiness. Indeed, besides loan underwriting and pricing, credit scores are
used extensively in the rental, labor, and auto insurance markets. For example, survey evidence suggests that up to 60 percent of employers, including the federal government, use
credit checks in their hiring decisions, while nearly all auto insurance providers take credit
record information into account in estimating the risk of car accidents (Chen, Corbae, and
Glover, 2013). Also, many cell phone and cable companies use credit score information in
contract-based plans. Furthermore, Dokko, Li, and Hayes (2015) document that the levels
and match quality of credit scores of a couple at the onset of their relationship have a pronounced predictive power regarding future relationship outcomes, even after controlling for
the credit events the couple encounters, such as new debt acquisitions and financial distress.
That said, it is important to remind ourselves that credit scores can be low for reasons that
8
For example, one of the credit reports prepared by the Retail Credit Company in 1934 included the
following questions: “Does his record show he has been a steady and reliable man?” “Is his personal
reputation as to character, honesty, and fair dealing good?” “Do you learn any illegal liquor traffic activities
or domestic difficulties?”
9
In addition, credit scoring also takes into account other factors, such as levels of indebtedness, length of
the credit history file, credit limit utilization, and public judgments, such as tax liens and wage garnishment
(Avery, Brevoort, and Canner, 2009).
10
have little to do with individuals’ general trustworthiness, and such factors are summarized
in the term of ability to repay, η. For example, many families with low credit scores have
gone through a negative financial event because of economic hardship brought on by severely
adverse events or job loss. The families with these negative financial experiences, though,
will be less likely to trust the advice of an institutional source (Alesina and La Ferrera, 2002;
Graeber and Zimmerman, 2016) or trust the financial market and institutions in general.10
To summarize, we argue that credit score is an indicator of one’s willingness and ability to repay financial debt and thereby contains information on an individual’s underlying
trustworthiness and previous experience with and attitudes toward financial markets and institutions. Accordingly, a community’s average credit score reveals information on the levels
of both trustworthiness and trust of its residents. Trustworthiness and trust in a community may also interact with and reinforce each other. For example, we can imagine that an
investor interacting with more trustworthy individuals in her community may grow to trust
other people more overtime. Community average credit scores therefore serve as a sound
indicator of the community’s social capital in general—a proposition that we test later in
the paper.
3
Data Description and Summary Statistics
Our study takes advantage of a rich array of data sources. Our proposed measure of social
capital—the average credit score of a community—is estimated using the Federal Reserve
Bank of New York Consumer Credit Panel/Equifax (henceforth FRBNY/Equifax CCP)
data. To establish this statistic as a valid indicator of social capital, we compare it with
a host of measures of trust and social capital used in the existing literature that include
the following: (1) self-reported trusting attitude in the 2000 Social Capital Community
Benchmark Survey (SCCS); (2) electoral participation in the 2012 U.S. general election by
10
SCF survey data suggest, for example, consumers who are assessed by the survey interviewers as more
suspicious to the survey interviews are indeed less likely to rely on advice from financial institutions but
more likely to stick with advice from friends, family, or themselves.
11
county from the Guardian newspaper; (3) blood donation data provided by the American
Red Cross; and (4) the number of consumer complaint filings at the FCC.
For information on household stock investment we use the Survey of Consumer Finances
(SCF) and the Panel Study of Income Dynamics (PSID). We are able to use the internal
version of the SCF and restricted geo-coded version of the PSID, so both datasets can be
linked to measures of social capital of a community. In order to control for community characteristics other than social capital, such as demographic compositions and contemporaneous
economic conditions, we use data from the Census Bureau, the Bureau of Labor Statistics
(BLS), the American Community Survey, the Federal Bureau of Investigation (FBI), and
CoreLogic. Finally, to control for local stock ownership, we take advantage of the Zip Code
Statistics of Income (SOI) data released by the Internal Revenue Service. This section will
introduce the primary data sources we use and present statistics of the key variables of our
study.
3.1
The FRBNY/Equifax CCP Data
The FRBNY/Equifax CCP is a large proprietary dataset that follows a 5-percent random
sample of U.S. consumers with valid credit histories (about 11 million individuals in recent
quarters) on a quarterly basis. The data include fairly detailed consumer residence location
information down to the census block level and extensive credit history data, including a
credit score. We calculate the average credit score at both the census tract and county levels.
A census tract has an average population of 4,000 of all ages. Relative to earlier studies on
the subject, which typically focus on cross-country or cross-province variations, working with
much smaller communities like census tracts allows us to measure social capital and analyze
its influence in a more zoomed-in, focused way. Our analysis removes the census tracts that
have fewer than 20 individuals in the sample.11 We compute the average credit score for
each quarter from the first quarter of 2001 to the fourth quarter of 2015. We then average
11
Because the FRBNY/Equifax CCP is a 5 percent random sample, we are roughly removing census tracts
with a population smaller than 400.
12
the quarterly mean of credit scores of each tract and county to filter out high-frequency
variations that are not necessarily reflecting differences in social trust and social capital. As
shown in the top panel of table 1, over the 2001–15 period, our sample has more than 655
million observations of individual credit scores, with a mean of 690 and a standard deviation
of 107. The standard deviations of the tract and county level average scores are 41 and
46, respectively, which are smaller than that of the individual score distribution but remain
quite sizable, suggesting wide geographical variations in community average credit scores
across the country.12
3.2
Other Measures of Trust and Social Capital
The 2000 SCCS samples 375 to 1,500 adults in 41 communities and asks the respondents
trust-related questions such as “whether most people can be trusted or you can’t be too
careful.” These trust-related questions greatly overlap with those from the World Value
Survey and General Social Survey, which are used extensively in studies of trust and trustworthiness. Each community in the survey is a county or a combination of several counties,
and we are able to map 38 out of the 41 communities to proper counties. We construct a
trust index that is equal to 100 times the share of survey respondents in a community who
reply that “most people can be trusted.” As shown in the lower panel of table 1, averaging
across all communities, 47 percent of the SCCS respondents in a community indicated that
they feel most people can be trusted, with a standard deviation of 8 percent.13
In addition, we collect data of electoral participation and blood donations as indicators
of social capital, which are introduced in GSZ (2004). Specifically, we use county-level 2012
presidential election participation data that cover over 2,700 U.S. counties. As shown in the
lower panel of table1, averaging across all counties, 42 percent of the total population in a
12
In addition, not shown in the table, standard deviations of the residuals of regressing the tract and county
level average scores on their respective socioeconomic and demographic characteristics remain sizable.
13
The 2000 SCCS data used in this study were collected by Professor Robert D. Putnam of the Saguaro
Seminar Civic Engagement in America, a project of the Kennedy School of Government at Harvard University
and numerous community foundations nation-wide, and made available through the Roper Center for Public
Opinion Research.
13
county voted in the 2012 general election, with a standard deviation of 8 percent.14
For blood donations, we use data provided by the American Red Cross. In 2015, the
American Red Cross collected blood from more than 13,300 Zip codes in more than 2,000
counties. These Zip codes cover more than 60 percent of the U.S. population. For consistent
comparisons, we aggregate the data to the county level. As shown in the lower panel of table
1, on average, 3.6 units of blood were collected per 100 people, with a standard deviation of
3.2 units.
Finally, we use the number of FCC complaints as additional (negative) indicators of
trust and social capital in an area.15 Similarly, we aggregate the number of complaints at
the county level. Statistics regarding the number of complaints per 100 population in each
county are presented in the lower panel of table 1. On average, about 0.14 complaints were
filed to the FCC per 100 residents, with considerable cross-county variations.
3.3
Survey of Consumer Finances
Our main focus is household stock investment decisions. Two large U.S. household surveys—
the SCF and PSID—provide such data along with detailed household level socioeconomic,
demographic, and financial characteristics, along with qualitative reports on preferences. Our
data include information on the census tract and county where sample households reside,
thereby allowing a merge of individual household investment decisions with measures of the
community’s social capital and other socioeconomic and demographic characteristics.
The SCF is conducted by the Federal Reserve Board every three years and is widely
regarded as one of the most comprehensive sources of data concerning U.S. household balance
sheets. These data have information on stock market participation status and share of stocks
in financial portfolio. We use three waves of the SCF data collected in 2004, 2007, and 2010.
During these years, the survey sample was stratified using the 2000 U.S. decennial census.
14
Note that 42 percent is computed using total population in each county as the denominator and total
voters in each county as the numerator. Because about 25 percent of the population is under 18, our estimate
of total national eligible voter turnout is 57 percent, which matches the national share.
15
The FCC filing data can be accessed at https://www.fcc.gov/consumer-help-center-data
14
Consumer location information in the FRBNY/Equifax CCP data is also coded using the
2000 census, ensuring a higher-quality match. During the sample period, as presented in
the left column of the top panel of table 2, only a small fraction (23 percent) of households
own corporate equities (stocks).16 Among those who own stocks, the share of stocks in their
financial assets portfolio is about 11 percent, with a fairly large dispersion in the sample.
In addition, the mean of wealth in the pooled 2004–10 SCF is about $437,000 in 2003
dollars, and mean income is about $69,000(not shown). Both are comparable with external
aggregates.17
The SCF also collects information on attitudes toward taking financial risk and investing.
Consistent with the apparently low levels of stock ownership, most families also report being
unwilling to take financial risk (table 2).18 Moreover, the SCF data are collected by a
trained field interviewer, either in person or on the phone. At the end of each completed
interview, the interviewer assesses how the respondent interacted with the survey. Among
these questions, the interviewer evaluates how suspicious the respondent was about the
survey before the interview began. We use this interviewer assessment to identify families
that were “not at all suspicious” of the interview, effectively creating a proxy for trusting
attitudes. Note that, in contrast to the World Value Survey or SCCS measures, this measure
of trusting attitude in the SCF is not self-reported but interviewer assessed. We include this
interviewer-assessed measure of trust later in the regression analysis for robustness. About
56 percent of SCF families are coded as “trusting” by the field interviewer.
16
Another 28 percent of families only own equities indirectly through tax-preferred retirement accounts.
At its broadest definition, then, equity ownership in the U.S. is slightly above 50 percent.
17
See Bricker, Henriques, Krimmel, and Sabelhaus (2016) for a comparison of SCF income and wealth estimates relative to those from income tax data. See Dettling and others (2015) for a more general comparison
of SCF aggregates with external sources.
18
Families that are “willing to take financial risks” are those that are either willing to take substantial or
above average financial risk when making investments. Families willing to take average financial risks and
those unwilling to take any financial risks are counted as not willing to take risk.
15
3.4
Panel Survey of Income and Dynamics
The PSID, unlike the SCF, is a longitudinal survey. It follows a core sample of households
and their offspring over nearly 50 years.19 From the 1999 survey, the PSID routinely collects
some basic household financial information, including stock and checking account ownership
and values. The PSID stock ownership is defined as including mutual funds but not including
IRAs and retirement accounts. We use the PSID data for both cross-sectional analysis of
stock ownership and dynamic analysis of stock market entries and exit.20
In the statistics presented in the right column of table 2, we note that the PSID stock
ownership is quite similar to that of the SCF. However, the stock share in financial assets is
higher in the PSID, likely due to the less-complete coverage of financial assets in the PSID
than in the SCF. In addition, about 6 percent of the households in the PSID sample that did
not own stock in a given year became stock investors two years later, and about 23 percent
of stock investors in a given year owned no stocks two years later. Moreover, regarding
attitudes toward financial risks, the PSID provides a risk tolerance index estimated using
lottery questions similar to those in Barsky, Juster, Kimball, and Shapiro (1997). A greater
value of the risk tolerance index indicates lower risk aversion, and the sample average is
slightly above 1. Because these questions were only included in the survey in 1996, the
tolerance index is available for only a fraction of the sample. Finally, 24 percent of the
households relocated to a different census tract within two years, and we focus on these
households in our stock market entry and exit analysis.
19
The PSID was an annual survey from 1968 to 1997, biennial afterwards.
For the cross-sectional analysis, we keep the first observation of a household in the longitudinal panel
that shares the same combination of location (census tract or county) and stock market participation status.
Therefore, the PSID sample constructed for cross sectional analysis is a somewhat younger than the SCF
sample. Some of the PSID data used in this analysis are derived from Restricted Data Files, obtained
under special contractual arrangements designed to protect the anonymity of respondents. These data are
not available from the authors. Persons interested in obtaining PSID Restricted Data Files should contact
[email protected].
20
16
3.5
Other Data Sources
In addition to individual characteristics, we also take into account potential effects of community characteristics other than trust and social capital on stock investment decisions. For
community demographic compositions, we use statistics of the 2000 U.S. Decennial Census
that include median income and racial, education, and age compositions. For local economic
conditions, we use the Bureau of Labor Statistics unemployment rates, the CoreLogic data
of house price growth, and the American Community Survey’s income inequality measurements, all available at the county level. In addition, we use the FBI’s violent crime rate
statistics in our social capital measurement analysis. We also use SOI Zip code-level data
on dividend income as a proxy of local average stock ownership. These data are released
by the Internal Revenue Service, and are computed as the ratio of the number of filers with
ordinary dividend income (line 9a of Form 1040) to the total number of filers in a given Zip
code.21 We use the public releases of the 2004, 2007, and 2010 SOI Zip code files, and match
these data to the SCF and PSID data of the closest waves.
4
Average Credit Scores as an Indicator of Social Capital
In this section, we implement a sequence of statistical analyses to validate average credit
score as an indicator for social capital of a community and discuss the strength and appeal
of this measurement.
4.1
Simple Correlations
To begin with, we show that average credit scores are correlated with a range of other measures of trust and social capital used in the literature—a survey-based trust indicator, blood
donations, electoral participation, and fraud complaints—even after taking into account a
21
There are almost certainly more filers in a Zip code than families (Bricker and others, 2016), but these
data can give a useful first approximation of the family-level statistics.
17
host of community characteristics. Specifically, we estimate the following model:
Indicatorcs = αs + β s Scorec + γ s Qc + εsc ,
(2)
where Indicatorcs is social capital indicator s in county c. Score is the county average credit
score, and Z is a vector of county level characteristics that includes the inverse hyperbolicsine transformation (I.H.S.) of median income, Gini coefficient of income, homeownership,
population shares of various educational attainments, share of age 65 and above, a measure
of racial diversity, and violent crime rates. The estimated coefficients of these regressions
are presented in table 3.
First, we note that the correlation between average credit scores and each of the four
measures of social capital and trust is economically and statistically significant, even after
controlling for a range of characteristics of the community. Specifically, our estimates indicate that counties with higher credit scores, on average, have a larger share of trusting
residents, who are more likely to participate in elections and blood drives, and file fewer
FCC complaints.
Putting the estimates in perspective, consider two otherwise comparable counties H
and L, where county H has an average credit score 46 points (one standard deviation)
higher than county L. Our estimates imply that in county H, 12 percentage points more
of its residents would believe “most people can be trusted,” 3 percentage points more of
its residents would vote in general elections, 3.7 more unites of blood (about the all-county
mean) would be collected, and 0.05 fewer FCC complaints per 100 residents (about half of
the all-county mean) would be filed. The negative relationship between average credit score
and the prevalence of FCC complaints likely indicate that communities with higher social
capital are less prone to be the targets of telemarketing scams, financial frauds, and harsh
debt collection treatments, and that residents in these communities are less suspicious and
are more alerted, and therefore more trustful, to certain market activities.22
22
See Raval (2016) for a discussion of the factors influencing the count of complaints to the Federal Trade
Commission.
18
Further, the explanatory power of the model in equation (2) increases when average credit
score (Score) is included. For self-reported trustfulness in the SCCS and blood donations,
adding average credit scores to the model boosts R-squared by a large margin (12 and 33
percent, respectively); for general election participation and FCC complaints, the increases in
R-squared are a more modest 2 to 3 percent (“memo” row of table 3). Finally, the uniformly
significant relationships between average credit scores and other social capital indicators
are even more notable when considering that none of the county characteristics included
as control variables (such as median income, inequality, education levels, racial diversity,
and violent crime rates) have consistent statistical relationships with these alternate social
capital indicators (columns 1–4 of table 3).23
4.2
Consistency among Social Capital Indicators
As a further test of consistency among the social capital indicators we consider, including
average credit scores, we run a series of regressions whereby each social capital indicator is
regressed against the others, with the same set of control variables as in equation (2).24 We
run these regressions using a subset of counties where all four social capital indicators are
available. The results are reported in the upper panel of table 4.
There are two notable observations in our estimates. First, consistent with our expectations, average credit scores, electoral participation, and blood donations are positively
correlated with each other, though the estimated coefficients are not all statistically significant. By contrast, FCC complaints are negatively correlated with the other three indicators.
Second, county average credit scores are almost always associated with the other social capital indicators in a fashion that is statistically significant and consistent with theoretical
predictions, underscoring the statistical richness and robustness of this indicator.25
23
Even without controlling for average credit scores, the statistical associations between the other social
capital indicators and the control variables remain noticeably weaker than that for average credit scores.
24
We omit the survey-reported SCCS trusting attitude for which we only have observations of 38 counties.
25
The only exception is in column 2, the coefficient is economically large but imprecisely estimated.
19
4.3
Principal Component Analysis
To the extent that they behave in a way consistent with our priors, each of our four social
capital indicators should convey information about a communitys social capital. Social
capital, though, is best interpreted as a latent variable. One common way of illustrating
this latent factor is through a principal component analysis (PCA). Here, we implement
a PCA of these four social capital indicators to underscore that average credit score is a
valid indicator of social capital and to illustrate that it accounts for the greatest amount of
variation in the latent factor among all indicators.
The PCA allows us to use our four observed county-level (c) social capital indicator
variables to model n(< 4) traits (Yn ):
Y1 = ε11 Scorec + ε21 DonateBloodc + ε31 V otingc + ε41 Complaintsc ,
Y2 = ε12 Scorec + ε22 DonateBloodc + ε32 V otingc + ε42 Complaintsc ,
(3)
...,
The first principal component is Y1 , the second principal component is Y2 , and so forth. As
shown in equation (3), the principal components are a linear combination of observed social
capital variables and estimated εnk s.26 Each principal component is, by design, uncorrelated
with the other principal components.
The principal components are not defined a priori but each is interpreted based on
commonalities in the variables that predict the component. As seen in the lower panel of table
4, credit scores, electoral voting, and blood donation are all highly correlated with the first
principal component, indicating that they are all signaling a similar trait.27 Notably, among
all the potential social capital measures, average credit score has the highest correlation with
the first principal component (column 1, lower panel).
26
The ε11 to εk1 are estimated by maximum likelihood. Essentially, the PCA finds the variable or variables
that explain the most variation in principal component Y1 —these variables have the largest values of ε1 —
then finds the variables that explain the most variation in principal component Y2 , and so forth (Jolliffe,
2002).
27
The first principal component explains about 45 percent of the variation in all the variables.
20
Because these principal components are derived from four series, a relatively small number, the correlation between each of the series and the principal components should be
interpreted with caution. To address this concern, we calculate the principal components
using three of the four social capital indicators and examine their correlations with the excluded series. The idea is that if a principal component derived from three indicators does
a good job in capturing underlying variations in social capital, this latent factor should be
correlated with the other indicators of social capital. Indeed, we find that average credit
score has the highest correlation with the first principal components derived from the other
three social capital measures (columns 3 and 4, lower panel), further reassuring its merit
as a valid indicator of social capital that is potentially richer and more robust than other
indicators exploited in the literature.
5
Static Analyses on Stock Ownership
As illustrated in the previous section , communities with higher average credit scores tend
to have a greater share of residents who are more trusting and a higher level of social
capital (measured with various indicators). We now revisit the relationship between social
capital and household financial decisions with an analysis of whether people living in such
high average credit score areas are more likely to own stocks. We begin with estimating a
workhorse model used extensively in stock market participation research, augmented with an
array of community characteristics, including average credit scores. In our baseline analysis,
a community is defined as a census tract. As previously discussed, an attractive feature
of using average credit scores to measure trust and social capital is that such an indicator
can be constructed for much more granular communities than earlier research. We also
estimate the model at the county level as a robustness check and to facilitate comparisons
with the estimates of other social capital indicators. Specifically, we estimate the following
logit model:
y
P artyi,t = α + βScoret + γZiy + θQt + ρY eary + εyi,t ,
21
(4)
where P artyi,t is a zero-one indicator of stock ownership (directly held or in mutual funds)
y
for household i that lives in county t in year y. Scoret is the mean credit score for county
t in year y. A positive β coefficient would suggest that residents living in communities
with higher credit scores are more likely to invest in stocks. Z is a vector of individual
characteristics of the investor, which includes the inverse hyperbolic sine transformation of
household income and wealth, a household head age polynomial, bins of head educational
attainment, race, marital status, and a single male dummy.28 Qt is a vector of county-level
community characteristics (median income, share of people with a college degree, share of
people with lower education, defined as high school or below, and share of white residents)
that may also affect investment decisions. Y ear is a vector of yearly fixed effects.
5.1
The SCF Analysis
We first estimate the model using the SCF data and the results are presented in table 5;
standard errors are clustered at the tract level and adjusted for multiple imputations in
the SCF shown in parentheses. In addition, for parameters of key interest, we also report
in brackets the implied odds ratio associated with a one-standard deviation change of the
independent variables.
To begin with, the baseline estimates shown in column 1 suggest that higher levels of
tract average credit scores are associated with greater stock ownership, and the relationship
is statistically significant and economically appreciable—a finding that is consistent with
earlier research. The estimated odds ratio indicates that residents living in a tract with an
average credit score 41 points (one standard deviation of the average credit score distribution
across all communities) higher than an otherwise identical tract are nearly 30 percent more
likely to own directly held equities in a given year, a margin similar to the estimates reported
in GSZ (2004).
28
The household characteristics included in Z are very similar to the existing literature on stock market
participation (see, for example, Haliassos and Bertaut, 1995; Campbell, 2006). We use the inverse hyperbolic sine transformation of household income and wealth to deal with zero and negative values. This
transformation is otherwise very similar to log transformation for typical positive values (Pence, 2006).
22
In addition, our estimated coefficients of the control variables are all statistically significant and mostly consistent with results reported in earlier research. For example, greater
levels of normal income and total wealth, greater willingness to take financial risk, and higher
educational attainments are all associated with greater stock ownership. In addition, not
shown in the table, none of the tract characteristics included as controls has a significant
positive relationship with equity ownership. We then conduct a series of extension and
robustness analysis to further corroborate the baseline results.
First, we show that the association between social capital and stock ownership in the
SCF is more pronounced among those with less education. As described in GSZ (2004,
2008), those with more education should have had more formal opportunities to understand
the benefits of investing in stocks, so in a world where trust determines stock investing, we
should see a larger effect for those with less education. That is exactly what we find when
we include terms that interact average credit scores with years of education (column 2). The
negative coefficient estimate on the interaction implies that the influence of social capital
is highest among those with fewer years of education (and, conversely, lower amongst those
with higher education levels).
Second, we exploit a unique feature of the internal SCF data: the interviewer’s assessment
of the individuals’ trusting attitude observed during data collection. As described earlier,
the interviewer who conducts the SCF in the field makes note of the responding family’s
degree of suspiciousness. Interestingly, there is a positive correlation between families rated
to have low suspicion and the average credit score of the tract where the respondent lives
(not shown), which is also consistent with the notion that people living in higher social
capital communities tend to more trustful. Consistent with the notion that more trusting
investors are more likely to invest in the stock market, the estimated odds ratio for the SCF
assessment of trust indicates that respondents who appeared to be more trusting of the survey
are on average 22 percent more likely to invest in stocks, a margin that is also statistically
significant. However, even after adding an alternative indicator of trustfulness to the model
23
as an additional control, our baseline results do not change qualitatively, and the average
credit score remains positively associated with stock ownership (column 3), underscoring the
information merit of credit scores that are orthogonal to other measures of trust and social
capital.
Third, if high credit score areas tend to have high stock ownership on average, then, instead of the effects of trust and social capital on stock investment, the estimated β coefficient
may reveal the effects of more efficient information sharing regarding equity investment that
leads to a higher probability of participation in the stock market for any given individual
investor (Hong, Kubik, and Stein, 2004; Li, 2014). To isolate the trust effect from this potential information sharing channel, we add to the model the local share of stock ownership
estimated using the SOI data. As shown in column 4, families living in areas that have
a one standard deviation higher stock ownership are 30 percent more likely to own stocks
themselves. Because an area’s average stock ownership is positively correlated with average
credit score, the estimated β coefficient is appreciably smaller in this specification, but the
β estimate is still economically sizable and statistically significant.
Including the local share of stock ownership in our cross-sectional estimates does not
fully address the potential endogeneity arising from local information sharing (as in Hong,
Kubik, and Stein, 2004, and Li, 2014). But it is re-assuring that the β estimate in column
4 is similar to another attempt at a causal relationship presented in table 5.
Fourth, we re-run our analysis using county, rather than census tract, as a community.
Our results (column 5) are qualitatively the same as those that use the tract as the community reference point (column 1). The smaller implied odds ratio for the county-level
analysis relative to the census tract level analysis (29 percent versus 22 percent) potentially
also speaks to the notion that the effect of social capital may be stronger for communities
defined at more granular levels, which in turn underscores this appealing feature of average
credit scores as such an indicator.29
29
Though not shown in the paper, we also examined whether our variable of interest was picking up credit
24
Finally, we study the effect of trust on the share of stock investment in household financial
asset portfolios by estimating a tobit model with the same controls as in column 1. The
results, reported in column 6, imply that a one standard deviation increase in tract mean
credit score is associated with a 7 percentage point increase in stock investment share (about
60 percent of the mean equity investment share, similar to GSZ (2008)).
5.2
The PSID Analysis
The PSID results, shown in table 6, are quite similar to the SCF results. Unlike the SCF,
which consistently collected risk aversion information in the three waves of data we use, the
PSID only asked for such information in a special module of 1996. As a result, we have risk
tolerance information for only a fraction of the households in the PSID sample. Therefore,
the PSID baseline regression, shown in column 1, does not include risk tolerance as a control.
The baseline PSID estimates show that households residing in census tracts that have a one
standard deviation higher average credit score, on average, are 24 percent more likely to
invest in stocks—similar to the SCF estimates. Notably, all control variables also have
rather similar coefficients estimated using the SCF and PSID data, adding confidence to the
representativeness of both surveys and the empirical model we adopt.
Interacting average credit scores with years of education of household heads in the PSID
sample (column 2) reveals that the association between our measure of social capital and
propensity of investing in stocks is diminishing with educational attainment—the same as
in the SCF analysis and consistent with GSZ (2004, 2008). Estimates in column 3 provide
reassurance that adding risk tolerance as a control variable does not qualitatively alter the
baseline results. While, as expected, investors with greater risk tolerance are more likely to
invest in stocks, in this model estimated using a smaller sample, greater community average
credit scores remain associated with higher propensity of owning stocks. Also, while local
average stock ownership is a powerful predictor of individual investors’ stock investment,
access in addition to social capital. However, our findings are the same qualitatively as in column (1) when
we re-run our analysis including a household-level measure of constrained credit as a control variable.
25
adding such a control does not void the baseline results—again, as in the SCF results (column
4). In addition, as in the SCF analysis, the results of census tract level analysis qualitatively
largely hold at the county level (column 5).
Moreover, the geo-coded PSID data include unique information about the county in which
one grew up. This information allows us to study how social capital of the community where
one grew up and the community where one currently lives may both influence an investor’s
stock ownership. To do so, we add the average credit score of the county in which the
household head grew up to the baseline model and estimate it using a sample of household
heads who no longer live in the same county in which they grew up. To highlight such a
contrast, we estimated the model using the gaps of average credit scores between the county
one grew up in and the county one currently lives in as the weights. The results, reported in
column 6 and consistent with GSZ (2004) and Brown, Ivković, Smith, and Weisbenner (2008),
show that higher social capital levels in both communities may boost household investors’
stock ownership, with the coefficient estimated for the grow-up county being substantially
larger.
There are two important caveats regarding this exercise. First, we note that we are
approximating the social capital one was exposed to while growing up (decades ago for some
investors) using that county’s current average credit score. While average credit score is
largely stable over time for most communities, it is possible that a county’s average credit
score today does not accuratgely reflect its social capital in the past. Second, the county one
relocated to is not necessarily independent of one’s earlier life experience, which in turn bears
the imprint of the county where one grew up. Thus, the social capital of the two counties
can be correlated, which needs to be taken into account while interpreting the coefficients
estimated from such a model. Finally, the PSID tobit regression results of stock share in a
financial portfolio (column 7) are very similar to those estimated using the SCF data.
26
5.3
Alternate Measures of Social Capital—A Horse-Race Test
We introduced community average credit scores as a measure of social capital in comparison
with other indicators used in the literature, including blood donations, electoral participation, and FCC complaints. Of these social capital indicators, blood donations and electoral
participation have been shown to have a positive effect on household financial decisions using
European data (GSZ, 2004). Here we estimate how various measures of social capital help
predict stock market participation using U.S. data and evaluate their respective significance
and robustness. We will first replace Score in equation (4) with each of the other three
social capital indicators. Then we estimate a variation of equation (4) that includes all four
indicators to assess their relative significance.
Following GSZ (2004) and because of the availability of electoral participation data, we
implement these analyses using county level average credit scores.30 Accordingly, in the
upper and lower panels of table 7, column 1 replicates the Score estimates in column 5 of
tables 5 and 6. A notable observation from the table is that electoral participation and
blood donations appear to have a significant bearing on stock ownership in the SCF or the
PSID. Moreover, fewer FCC complaints are associated with greater propensity of investing
in stocks only in the SCF sample. Finally, when all four social capital measures are included,
as shown in column 5, the community average credit score remains a prominent predictor of
stock market participation.
6
Dynamic Analysis of Stock Market Entries and Exits
While the results of cross-sectional analyses presented earlier are robust and strongly indicative regarding the potential effects of trust and social capital on household stock investment,
concerns remain regarding whether these results establish a causal relationship or are driven
by stock investors being more likely to live in high credit score areas, holding other factors
30
More precisely, GSZ (2004) use Italian provincial data. They note that provinces in Italy are similar to
U.S. counties.
27
constant. To address this concern, we follow Li (2014) and exploit the longitudinal structure
of the PSID and ask whether an investor who did not own stocks before will have a higher
chance of entering the stock market after moving to a community with a higher average
credit score. Specifically, for an investor who did not own stocks in year y − 2, moved to a
different community sometime between y − 2 and y, and still owned no stocks in year y, we
estimate the following logistic model of her probability of entering the stock market by year
y + 2.
entryiy, y+2 = α+βb CS ty−2 +βp ∆p CS ty−2 ,ty +βn ∆n CS ty−2 ,ty +γZiy +θ∆Qty−2 ,ty +ρY eary +εyi,t ,
(5)
where entryiy,y+2 is an indicator of entering the stock market between year y and y + 2, and
ty−2 and ty denote the tract one resided in during year y −2 and y, respectively. Accordingly,
CS ty−2 denotes the average credit score of the census tract investor i resided in during year
y − 2. ∆p CS ty−2 ,ty and ∆n CS ty−2 ,ty are the positive or negative changes of average score
before and after the relocation, respectively, to allow for asymmetric effects on subsequent
stock market entry decisions. Control variables in Z are defined similarly as in equation (4),
but here we use the y − 2, y, and y + 2 average levels of wealth and income. In addition, we
include the change of real income between y − 2 and y + 2 to take into account the potential
effects of the factors that led to the relocation on stock market entry. Moreover, we control
for the changes of community characteristics ∆Qty−2 ,ty to address the potential effects on
stock ownership of these factors. Investors who did not move between years y − 2 and y are
included as the control group.
We focus on the stock market entry dynamics observed after the move in order to isolate
the stock market entries that are endogenous to the relocation decisions. Indeed, while we
control for an extensive set of indicators of household financial and demographic condition
changes, there might be unobserved factors that cause the household to decide to move to
a new neighborhood and start investing in stocks at the same time. Focusing on the stock
28
market entries after moving helps alleviate this endogeneity concern. Furthermore, similar
to Li (2014), we examine whether relocating to a lower credit score neighborhood increases
a current stock investor’s odds of subsequently exiting the stock market. For stock investors
in year t − 2 and y, we estimate a similar model of stock market exits between y and y + 2.
Arguably, current stock owners’ investment decisions (including exiting the stock market)
are more determined by their investment experiences, financial conditions, and expected
returns, but less influenced by trusting attitude changes.
The results are summarized in table 8. While the estimates in column 1 do not suggest
that moving to a community of higher average credit scores has a significant effect on subsequent stock market entries, this appears to reflect the asymmetry of the potential effects of
social capital. As shown in columns 2 and 3, an investor who did not own stock previously
and moved to a census tract of higher credit score would have a higher chance of entering the
stock market during the two years following the move. The estimated odds ratio suggests
that the relocation-induced positive change in community average credit scores that is one
standard deviation bigger implies an 11 percent higher chance of entering the stock market.31
By contrast, while the effects on the likelihood of entering the stock market are negative for
moving to communities with lower credit scores, they are not statistically significant and
much smaller in magnitude. Furthermore, our estimates, as in the cross-sectional analysis,
are not sensitive to the inclusion of changes in community average stock ownership as a control variable (column 4). Finally, as shown in columns 5–8, relocation-induced community
average credit score changes do not appear to have any significant effect on current stock
owners’ decisions on exiting the market. The contrast between the estimates of market entry
and exit, as we argued earlier, is consistent with the notion that current stock investors tend
to make investment decisions on objective, market related factors, and they are therefore
less influenced by subjective perceptions, such as trusting attitudes. Consequently, such a
31
On average, 6 percent of households that did not invest in stocks entered the market within a two-year
period.
29
contrast also lends additional support to a causal relationship between social capital and
stock investment.
7
Conclusion
This paper achieves two goals. First, we introduce average credit scores as a novel measure
of a community’s social capital. We show that this measure is consistent with other social
capital measures employed in previous research. Average credit score as a measure of social
capital is appealing for several reasons. It is objective, data driven, and based on individual
behavior. Various analysis we conduct in this paper suggest that this measure is richer and
more robust than other indicators. Such an indicator can be constructed for very small communities, including census tracts or even street blocks, and the underlying data are available
for essentially the entire country. It is more correlated with the latent factor derived from
a host of measures of social capital. Second, we revisit the relationship between social capital, trust, and investment in stocks using U.S. data. We find that while other measures of
social capital do not consistently show such a relationship, average community credit scores
consistently reveal that greater trust and social capital enhances stock market participation. Furthermore, such a relationship is more pronounced for lower-educated investors and
manifests itself in both static and dynamic analysis, all suggesting a causal relationship.
This new measure pushes the frontiers of social capital research. Many questions on this
subject that we were previously not able to answer due to measurement and data limitation
can be revisited. There are also many directions in which this measure can be enriched. For
example, the current indicator only focuses on the first moment of community credit score
distribution. We can imagine that the dispersion and tail properties of the distribution may
reveal new insights on the social capital of a community.
30
References
[1] Alesina, Alberto and Eliana La Ferrrara (2002) “Who Trusts Who?” Journal of Public
Economics, vol. 85, pp. 207–34.
[2] Algan, Yann and Pierre Cahuc (2010). “Inherited Trust and Growth,” American Economic Review, vol. 100 (December), pp. 2060–92.
[3] Arrow, Kenneth (1969) “The Organization of Economic Activity: Issues Pertinent to
the Choice of Market versus Nonmarket Allocation. The Analysis and Evaluation of
Public Expenditure: the PPB System”, A Compendium of Papers Submitted to the
Subcommittee on Economy in Government of the Joint Economic Committee, Congress
of the United States 1, pp. 47–64.
[4] Arrow, Kenneth (1972) “Gifts and Exchanges,” Philosophy and Public Affairs, vol. 1,
pp. 343–362.
[5] Avery, Robert, Kenneth Brevoort, and Glenn Canner (2009). “Credit Scoring and Its
Effects on the Availability and Affordability of Credit,” Journal of Consumer Affairs,
vol. 43 (September), pp. 516-37.
[6] Balloch, Adnan, Annamaria Nicolae, and Dennis Philip (2015). “Stock Market Literacy,
Trust and Participation”. Review of Finance, vol. 19, pp. 1925–63.
[7] Barsky, Robert, Thomas Juster, Miles Kimball, and Matthew Shapiro (1997). “Preference Parameters and Behavioral Heterogeneity: An Experimental Approach in the
Health and Retirement Study,” Quarterly Journal of Economics, vol. 112 (May), pp.
537–79.
[8] Bricker, Jesse, Lisa J. Dettling, Alice Henriques, Joanne W. Hsu, Kevin B. Moore, John
Sabelhaus, Jeffrey Thompson, and Richard A. Windle (2014). “Changes in U.S. Family
31
Finances from 2010 to 2013: Evidence from the Survey of Consumer Finances,” Federal
Reserve Bulletin, vol. 100 (September), pp. 1–41.
[9] Bricker, Jesse, Alice Henriques, Jacob Krimmel, and John Sabelhaus (2016) “Measuring
Income and Wealth at the Top Using Administrative and Survey Data,” Brookings
Papers on Economic Analysis, Spring.
[10] Briggs, Joseph, David Cesarini, Erik Lindqvist, and Robert ¨stling (2016) “Windfall
Gains and Stock Market Participation.” working paper.
[11] Brown, Jeffrey, Zoran Ivković, Paul Smith, and Scott Weisbenner (2008) “Neighbors
Matter: Causal Community Effects and Stock Market Participation,” Journal of Finance, vol. 63 (June), pp. 223–37.
[12] Campbell, John (2006) “Household Finance,” Journal of Finance, vol. 61 (August) pp.
1553–604.
[13] Chen, Daphne, Dean Corbae, and Andrew Glover (2013) “Can Employer Credit Checks
Create Poverty Traps? Theory and Policy.” working paper.
[14] Dettling, Lisa J., Sebastian J. Devlin-Foltz, Jacob Krimmel, Sarah J. Pack, and Jeffrey
P. Thompson (2015). “Comparing Micro and Macro Sources for Household Accounts
in the United States: Evidence from the Survey of Consumer Finances,” Finance and
Economics Discussion Series 2015–086. Washington: Board of Governors of the Federal
Reserve System, June.
[15] Dokko, Jane, Geng Li, and Jessica Hayes (2015). “Credit Scores and Committed Relationships,” Finance and Economics Discussion Series 2015–081. Washington: Board of
Governors of the Federal Reserve System, October.
32
[16] El-Attar, Mayssun and Markus Poschke (2011). “Trust and the Choice Between Housing
and Financial Assets: Evidence from Spanish Households,” Review of Finance, vol. 15
(October), pp. 727–56.
[17] Fukuyama, Francis (1995). “Trust: The Social Virtues and the Creation of Prosperity,”
New York: The Free Press.
[18] Duarte, Jefferson, Stephan Siegel, and Lance Young (2012). “Trust and Credit: The
Role of Appearance in Peer-to-Peer Lending”. Review of Financial Studies, vol. 25(8),
pp. 2455–83.
[19] Fehr, Ernst, Urs Fischbacher, Bernhard von Rosenbladt, Juergen Schupp, and Gert
Wagner (2003). “A Nation-Wide Laboratory. Examining Trust and Trustworthiness
by Integrating Behavioral Experiments into Representative Surveys,” CESifo Working
Paper Series No. 866.
[20] Georgarakos, Dimitris and Giacomo Pasini (2011). “Trust, Sociability and Stock Market
Participation,” Review of Finance, vol. 15 (October), pp. 693–725.
[21] Giannetti, Mariassunta and Tracy Wang (2016). “Corporate Scandals and Household
Stock Market Participation,” Journal of Finance, vol. 71 (December), pp. 2591–636.
[22] Glaeser, Edward, Javid laibson, Jose Scheinkman, and Christine Soutter (2000). “Measuring Trust” Quarterly Journal of Economics vol. 115 (August), pp. 811–46.
[23] Gennaioli, Nicola, Andrei Schleifer, and Robert Vishny (2015). “Money Doctors”, Journal of Finance, vol. 70 (February), pp. 91–114.
[24] Goldin, Claudia and Lawrence Katz (1999). “Human Capital and Social Capital: The
Rise of Secondary Schooling in America, 1910 to 1940,” Journal of Interdisciplinary
History, vol. 29, pp. 683–723.
33
[25] Graeber, Thomas and Tom Zimmerman (2016). “The Legacy of Financial Crises – A
Persistent Erosion of Interpersonal Trust?”, mimeo.
[26] Guiso, Luigi, Paola Sapienza, and Luigi Zingales (2004). “The Role of Social Capital in
Financial Development,” American Economic Review, vol. 94 (June), pp. 526–56.
[27] Guiso, Luigi, Paola Sapienza, and Luigi Zingales (2008). “Trust the Stock Market,”
Journal of Finance, vol. 63 (December), pp. 2557–600.
[28] Haliassos, Michael and Carol C. Bertaut (1995). “Why Do So Few Hold Stock?”, The
Economic Journal, vol. 105 (September), pp. 1110–29.
[29] Helliwell, John F., Haifang Huang, and Shun Wang (2016) “New Evidence on Trust and
Well-being”, NBER Working Paper No. 22450.
[30] Hong, Harrison, Jeffrey D. Kubik, and Jeremy C. Stein (2004), “Social Interaction and
Stock Market Participation,” The Journal of Finance, vol. 59 (February), pp. 137–63.
[31] Jolliffe, I.T. (2002). “Principal Component Analysis, 2nd ed.” New York: SpringerVerlag.
[32] Karlan, Dean (2002). “Using Experimental Economics to Measure Social Capital and
Predict Financial Decisions,” American Economic Review, vol. 83 (December), pp. 168899.
[33] Knack, Stephen and Philip Keefer (1997). “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation,” Quarterly Journal of Economics vol. 112 (November), pp. 1251–88.
[34] La Porta, Rafael, Florencio, Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny
(1997). “Trust in Large Organizations” American Economic Review vol. 87 (May), pp.
333–8.
34
[35] Li, Geng (2014). “Information Sharing and Stock Market Participation: Evidence from
Extended Families” Review of Economics and Statistics vol. 96 (February), pp. 151–60.
[36] Malmendier, Ulrike and Stefan Nagel (2011). “Depression Babies: Do Macroeconomic
Experiences Affect Risk Taking?” Quarterly Journal of Economics vol. 126 (February),
pp. 373–416.
[37] Malmendier, Ulrike and Stefan Nagel (forthcoming). “Learning from Inflation Experiences” Quarterly Journal of Economics forthcoming.
[38] Mankiw, Gregory and Stephen Zeldes (1991). “The Consumption of Stockholders and
Nonstockholders” Journal of Financial Economics vol. 29 (March), pp. 97–112.
[39] Pence, Karen (2006). “Foreclosing on Opportunity: State Laws and Mortgage Credit,”
Review of Economics and Statistics , vol. 88 (February), pp. 177–82.
[40] Putnam, Robert (1995). “Bowling Alone,” Journal of Democracy vol. 6 (January) pp.
65-78.
[41] Putnam, Robert (1993). Making Democracy Work: Civic Traditions in Modern Italy.
Princeton: Princeton University Press.
[42] Raval, Devesh (2016). “What Determines Consumer Complaining Behavior?” mimeo.
[43] Sapienza Paola, Anna Toldra-Simats, and Luigi Zingales (2013). “Understanding Trust,”
The Economic Journal, vol. 123 (June) pp. 1313–32.
[44] Vissing-Jorgensen, Annette (2002). “ Limited Asset Market Participation and the Elasticity of Intertemporal Substitution” Journal of Political Economy vol. 110 (August),
pp. 825-53.
35
Table 1: Summary Statistics of Credit Scores and Measures of Trust and Social Capital
Credit score data are from the FRBNY/Equifax CCP data. Statistics are estimated by pooling quarterly
data from 2001 to 2015 (60 quarters). Self-reported trust indicator is estimated using the Social Capital
Community Benchmark 2000 Survey. Electoral participation statistics are estimated from the November
2012 general election. Blood donations data are from the American Red Cross. County-level statistics are
aggregated from the Zip codes from which the Red Cross collected blood. The Federal Communications
Commission complaints data can be accessed at https://www.fcc.gov/consumer-help-center-data.
Credit Scores
Individuals
Census tract average
County average
Mean
690
684
680
S.D.
(107)
(41)
(46)
655 million
74,434
3,856
N
Other Social Trust and Social Capital Indicators
Self-reported trust
Electoral participation
Blood donations
FCC complaints
Units
Share reporting trust
Share voted
Units collected
per 100 residents
Complaints submitted
per 100 residents
Mean
47.1%
42%
3.6
0.14
S.D.
(8.1%)
(8%)
(3.2)
(0.14)
38
2,712
2,054
3,043
N
36
Table 2: Summary Statistics of Stock Investment and Household Characteristics
Statistics are estimated using the 2004, 2007, and 2010 Survey of Consumer Finances data and the 1999–2013
Panel Study of Income Dynamics data.
Variable
Financial decisions
Stockholder (%)
SCF
PSID
23.5
22.3
11.1
16.0
Enter stock market (%)
...
6.3
Exit stock market (%)
...
23.3
Rely on family/friends for fin. dec. (%)
20.0
...
Rely on formal source for fin. dec. (%)
32.8
...
18.8
...
...
1.1
56.1
...
...
24.3
Stock portfolio share (%)
Household characteristics
Willing to take fin. risk (%)
Risk tolerance
Interviewer observed trust (%)
Relocated in past two years (%)
37
Table 3: Average Credit Scores and Indicators of Trust and Social Capital
This table reports regressions of indicators of trust and social capital studied in the existing literature on county average credit scores. Standard
errors are clustered at the state level and presented in parentheses. *, **, and *** denote 90, 95, and 99 percent statistical significance, respectively.
Control variables include the inverse hyperbolic sine transformation of median income, income Gini coefficient, share of homeowners, composition
of educational attainment, share of senior population, Herfindahl index on racial diversity, and violent crime rate, all at the county level. The
“memo” row shows the percent increase of R-squared when adding average credit scores to the estimation. Column 5 projects average credit scores
on the control variables.
Self-reported trust
Electoral participation
Blood donations
FCC complaints
(1)
(2)
(3)
(4)
Score
100
(5)
0.267*
0.070*
8.054***
-0.094**
...
(0.141)
(0.041)
(2.217)
(0.044)
I.H.S. median income
-0.040
(0.150)
-0.051
(0.047)
-4.274*
(2.263)
0.184***
(0.059)
0.290***
(0.062)
Gini coeff.
-0.043
(0.039)
0.011*
(0.006)
-0.519
(0.357)
0.004
(0.012)
-0.041***
(0.009)
Homeownership
-0.316
(0.275)
0.110*
(0.059)
1.799
(2.194)
-0.141
(0.087)
-0.295***
(0.074)
High school and below share
-0.340
(0.466)
-0.351***
(0.103)
1.483
(4.276)
-0.082
(0.111)
-0.171
(0.134)
College graduate share
0.113
(0.342)
0.262***
(0.088)
-0.435
(4.008)
0.784***
(0.186)
1.277***
(0.129)
Senior population share
-0.921
(0.830)
0.292*
(0.159)
-0.149
(6.064)
0.071
(0.255)
1.782***
(0.293)
Racial Herf. index
0.345
(0.250)
-0.018
(0.058)
-0.987
(2.000)
0.087
(0.053)
0.751***
(0.068)
Violent crime rate
0.010
(0.013)
38
-0.003
(0.002)
2,571
-0.012
(0.080)
1970
0.001
(0.003)
2,898
-0.020***
(0.003)
2,973
11.7
2.8
33.3
2.1
...
Score
100
38
N
Memo
∆R-Squared (%)
Table 4: Consistency among All Social Capital Indicators
The upper panel of this table presents the consistency analysis among four indicators of social capital,
including the average credit score. Each of these social capital indicators is regressed on the other three
indicators and the set of control variables in table 3. Standard errors are clustered at the state level and
presented in parentheses. *, **, and *** denote 90, 95, and 99 percent statistical significance, respectively.
We found that average credit scores, blood donations, and electoral participation are positively correlated
among themselves, while all are negatively correlated with FCC complaints (though not all coefficients are
statistically significant). The lower panel shows that the average credit score has the highest correlation with
the first principal component extracted from the four indicators (column 1). Moreover, it has the highest
correlation with the first principal component extracted from the other three indicators (column 3).
Competing Regressions
Score
100
Electoral participation
Blood donations
FCC complaints
Controlled for
county chars.
N
Score
100
(1)
Electoral participation
Blood donations
FCC complaints
(2)
(3)
(4)
...
0.057
8.864***
-0.092**
(0.055)
...
(2.301)
3.584
(3.949)
...
(0.045)
-0.023
(0.092)
-0.003*
(0.002)
...
0.145
(0.145)
0.008***
(0.001)
-0.004
(0.016)
0.001
(0.001)
-0.040
(0.026)
-1.427
(1.172)
Yes
Yes
Yes
Yes
1,687
1,687
1,687
1,687
F irst P C 4
(1)
Correlations with Principal Components
Second P C 4
F irst P C 3
(2)
(3)
Second P C 3
(4)
Score
100
Electoral participation
0.85***
0.09***
0.53***
0.19***
0.73***
0.09***
0.39***
0.03
Blood donations
0.68***
-0.39***
0.33***
-0.18***
FCC complaints
0.13***
0.94***
0.05**
-0.12***
39
Table 5: Community Average Credit Scores and Stock Investment—SCF Analysis
Note: Standard errors are presented in parentheses. Odds ratios associated with a one standard deviation change of the key
independent variables are presented in brackets. Data are 2004, 2007, and 2010 SCF, pooled together. Credit score averages
are calculated using the FRBNY/Equifax CCP data. Standard errors are clustered at the census tract and county level,
respectively, and corrected for multiple imputation. *, **, and *** denote 90, 95, and 99 percent statistical significance,
respectively. Column 1 presents the results of the baseline specification. Column 2 tests if the coefficient estimated for average
credit scores diminishes with years of schooling. Columns 3 and 4 test the robustness of baseline results when adding SCFinterviewer-observed trustfulness and neighborhood average stock ownership as controls. Column 5 estimates the baseline model
using county level data. Column 6 studies the association between average credit scores and stock shares in household portfolios
of financial assets.
Logistic Analysis of Stock Ownership
Tract level
Score
100
(3)
(4)
(5)
(6)
0.621**
1.582**
0.630**
0.321**
0.861**
0.173**
(0.137)
[1.293]
(0.515)
[1.925]
(0.135)
[1.291]
(0.147)
[1.136]
(0.212)
[1.223]
(0.041)
0.633**
(0.060)
0.097**
(0.013)
-0.046**
(0.011)
0.001**
(0.000)
0.207**
(0.013)
0.348**
(0.115)
0.147**
(0.087)
0.619**
(0.082)
0.581**
(0.067)
0.127**
(0.015)
0.024**
(0.003)
-0.007**
(0.003)
0.000**
(0.000)
0.056**
(0.004)
0.132**
(0.033)
0.067**
(0.030)
0.170**
(0.023)
0.185**
(0.019)
Yes
Yes
Yes
Yes
14,137
Yes
Yes
Yes
Yes
14,122
-0.067*
(0.034)
[0.972]
0.201**
(0.056)
[1.223]
Zip code stock ownership
Head age
Head age2
Yrs. ed.
Married
Single male
White
Willing to take fin. risk
Controlled for
Family size dummies
Tract characteristics
County economic conditions
Yearly fixed effects
N
Tract level
(2)
SCF observed trusting
I.H.S. real wealth ($2002)
County level
(1)
Score
× Yrs. ed.
100
I.H.S. real income ($2002)
Tobit Analysis
0.560**
(0.053)
0.092**
(0.012)
-0.041**
(0.011)
0.000**
(0.000)
0.188**
(0.015)
0.357**
(0.112)
0.176*
(0.099)
0.534**
(0.081)
0.592**
(0.066)
0.561**
(0.053)
0.092**
(0.012)
-0.040**
(0.011)
0.000**
(0.000)
0.658**
(0.241)
0.358**
(0.112)
0.178*
(0.099)
0.530**
(0.082)
0.590**
(0.066)
0.562**
(0.053)
0.092**
(0.012)
-0.039**
(0.011)
0.000**
(0.000)
0.186**
(0.014)
0.362**
(0.112)
0.164*
(0.099)
0.532**
(0.081)
0.589**
(0.066)
2.123**
(0.387)
[1.304]
0.562**
(0.053)
0.091**
(0.012)
-0.040**
(0.011)
0.000**
(0.000)
0.188**
(0.015)
0.365**
(0.113)
0.183*
(0.099)
0.524**
(0.082)
0.594**
(0.066)
Yes
Yes
Yes
Yes
14,122
Yes
Yes
Yes
Yes
14,122
Yes
Yes
Yes
Yes
14,122
Yes
Yes
Yes
Yes
14,113
40
Table 6: Community Average Credit Scores and Stock Ownership—PSID Analysis
Note: Standard errors are presented in parentheses. Odds ratios associated with a one standard deviation change of the key
independent variables are presented in brackets. Data are 1999–2013 PSID (see the text for sample construction details).
Credit score averages are calculated using the FRBNY/Equifax CCP data. Standard errors are clustered at the census tract
and county level, respectively. *, **, and *** denote 90, 95, and 99 percent statistical significance, respectively. Column 1
presents the results of the baseline specification. Column 2 tests if the coefficient estimated for average credit scores diminishes
with years of schooling. Columns 3 and 4 test the robustness of baseline results when adding indicators of risk aversion and
neighborhood average stock ownership as controls. Column 5 estimates the baseline model using county-level data. Column
6 contrasts the estimates of the average credit scores of the current-residence county and the grow-up county. This model is
estimated using the gaps of average credit scores between the county one grew up in and the county one currently lives in as the
weights. Column 7 studies the association between average credit scores and stock share in the household portfolio of financial
assets.
Logistic Analysis of Stock Ownership
(1)
Score
100
County grew up
Tract level
(2)
(3)
(4)
County level
(5)
(6)
2.358***
0.307**
0.322***
0.482***
0.139***
0.128***
(0.093)
[1.245]
(0.360)
[2.921]
(0.135)
[1.147]
(0.101)
[1.158]
(0.169)
[1.116]
(0.032)
[1.179]
(0.040)
0.362***
(0.023)
[1.599]
Score
× years of edu.
100
-0.135***
(0.024)
Risk tolerance
0.032***
(0.011)
ZIP code stock ownership
I.H.S. real wealth ($2002)
Head age
Head age2
Yrs. ed.
Married
Single male
White
Controlled for
Family size dummies
Tract characteristics
County economic conditions
Yearly fixed effects
N
Tract level
(7)
0.481***
Score
100
I.H.S. real income ($2002)
Tobit Analysis
1.045***
(0.235)
0.612***
(0.036)
0.070***
(0.005)
-0.005
(0.006)
0.019***
(0.006)
0.184***
(0.009)
0.218***
(0.057)
0.082
(0.059)
0.666***
(0.055)
0.605***
(0.036)
0.071***
(0.005)
-0.004
(0.006)
0.018***
(0.006)
1.117***
(0.171)
0.222***
(0.057)
0.084
(0.059)
0.657***
(0.055)
0.566***
(0.060)
0.097***
(0.012)
-0.008
(0.014)
0.020
(0.014)
0.130***
(0.013)
0.306***
(0.096)
0.047
(0.096)
0.610***
(0.085)
0.612***
(0.037)
0.072***
(0.005)
-0.004
(0.006)
0.017***
(0.006)
0.183***
(0.009)
0.208***
(0.057)
0.083
(0.060)
0.666***
(0.056)
0.630***
(0.037)
0.063***
(0.004)
0.007
(0.007)
0.008
(0.007)
0.170***
(0.009)
0.254***
(0.058)
0.130**
(0.062)
0.766***
(0.059)
0.665***
(0.010)
0.055***
(0.001)
-0.004**
(0.002)
0.016***
(0.002)
0.186***
(0.003)
0.328***
(0.018)
0.152***
(0.019)
0.840***
(0.016)
0.186***
(0.015)
0.023***
(0.002)
0.005*
(0.003)
0.002
(0.003)
0.065***
(0.004)
0.069***
(0.025)
0.083***
(0.026)
0.233***
(0.024)
Yes
Yes
Yes
Yes
22,731
Yes
Yes
Yes
Yes
22,731
Yes
Yes
Yes
Yes
7,064
Yes
Yes
Yes
Yes
22,396
Yes
Yes
Yes
Yes
18,073
Yes
Yes
Yes
Yes
10,653
Yes
Yes
Yes
Yes
16,587
41
Table 7: Other Measures of Social Capital and Stock Ownership
Note: Standard errors are presented in parentheses. Data used in estimation of the upper and lower panel
are 2004, 2007, and 2010 SCF, pooled together, and a sample constructed using the 1999–2013 PSID,
respectively. Control variables are the same as in tables 5 and 6. Credit score averages are calculated
using the FRBNY/Equifax CCP data, county voting data are from the Guardian newspaper, county blood
donation data are estimated using data from the American Red Cross, and FCC complaints data can be
accessed at https://www.fcc.gov/consumer-help-center-data. Standard errors are clustered at county level
and corrected for multiple imputation (SCF only). *, **, and *** denote 90, 95, and 99 percent statistical
significance, respectively.
SCF
(1)
Score
100
(2)
(3)
(4)
0.861**
0.607**
(0.212)
Pct. voting
(5)
-0.351**
(0.151)
(0.273)
0.752
(0.682)
0.023
(0.016)
-0.161**
(0.078)
(4)
(5)
0.798
(0.584)
Pct. donating blood
0.016
(0.015)
FCC complaints
PSID
(1)
Score
100
Pct. voting
(2)
(3)
0.481***
0.663***
(0.169)
0.142
(0.380)
Pct. donating blood
-0.008
(0.009)
FCC complaints
0.056
(0.131)
42
(0.225)
0.051
(0.425)
-0.004
(0.0010)
0.122
(0.141)
Table 8: Dynamic Analysis
Note: Standard errors are presented in parentheses. Odds ratios associated with a one standard deviation change of the independent variables
are presented in brackets. Data are 1999–2013 PSID (see the text for sample construction details). Credit score averages are calculated using
the FRBNY/Equifax CCP data. Standard errors are clustered at the census tract level. *, **, and *** denote 90, 95, and 99 percent statistical
significance, respectively. Columns 1–4 present the results of how, for those who do not own stocks, relocating to a community of different average
credit score may affect the odds of entering the stock market in the years after the move. Columns 5–8 present the results of how, for those who
currently own stocks, relocating to a community of different average credit score may affect the odds of exiting the stock market in the years after
the move.
Entry analysis
(1)
∆
Score
100
43
∆+
∆−
(4)
(5)
0.150
(0.252)
[1.046]
(0.270)
[1.020]
Score
100
Controlled for
Individual char.
∆ Local char. and econ conditions
∆ Zip code stock ownership
Yearly fixed effects
N
(3)
0.272
Score
100
CS 0
100
(2)
Exit analysis
(6)
(7)
(8)
0.517*
0.885**
0.871**
0.272
-0.214
-0.291
(0.300)
[1.064]
(0.383)
[1.111]
(0.394)
[1.110]
(0.333)
[1.029]
(0.432)
[0.978]
(0.447)
[0.970]
-0.074
-0.427
-0.407
-0.087
0.877
0.998
(0.340)
[0.992]
(0.419)
[0.956]
(0.426)
[0.958]
(0.447)
[0.993]
(0.596)
[1.072]
(0.615)
[1.082]
0.457***
0.460***
0.456***
0.458***
-0.374***
-0.370***
-0.369***
-0.355***
(0.110)
1.222
(0.110)
[1.224]
(0.110)
[1.222]
(0.110)
[1.222]
(0.127)
[0.885]
(0.127)
[0.885]
(0.127)
[0.886]
(0.127)
[0.890]
Yes
No
No
Yes
19,880
Yes
No
No
Yes
19,880
Yes
Yes
No
Yes
19,880
Yes
Yes
Yes
Yes
19,880
Yes
No
No
Yes
4,539
Yes
No
No
Yes
4,539
Yes
Yes
No
Yes
4,539
Yes
Yes
Yes
Yes
4,539